Ideal Core (Core Parking Optimization) and how to change it!

[EastCoasthandle]

I believe this deserves it's own thread. I stumbled upon this while researching something else and thought this may be benefit for others. From what I've gathered Core Parking moves all processing to either a core or cores and putting the other cores to sleep. Here is what I found on the subject:

If a task’s needs are being addressed by one core, the operating system will let you stay there. This means two things to Intel: first, you don’t use power on the migration, and second, idle cores are able to remain in a C6 state. Purportedly, this migration fix alone will yield an extra 10 to 15 minutes of battery life on Nehalem-based notebooks, though this won’t become a major issue until the mobile dual-core Arrandale launches later this year. Perhaps more interesting, though, is that processors without C6 will not realize this gain (including AMD’s CPUs).

Core parking is a second optimization, based on the observation in previous operating systems that you might have four cores running background processes at 10% utilization each. The idea is to load all of those tasks onto onecore and let the others idle if operating load levels allow for it. Now, you can see how these two features working together might have a significant impact on power, as ideal core prevents rabid thread migration, while core parking optimizes loading. Taken together, the pair intelligently maximizes the number of idle cores, and then keeps them from being spun up unnecessarily, yielding the theoretical power gains.

The results were actually counter to what we expected. The Windows 7-based build averaged six watts higher over the course of its run, but finished the test three minutes faster than the Vista machine. Also noteworthy, though, is that when theWindows 7 machine has a chance to idle (which is where we'd expect to see ideal core and core parking actually having an effect), it does dip down lower than the Windows Vista box.

We checked these results over with Intel, and came away with the following interpretation: the Windows 7 P-state promotion policies are more aggressive than Vista's, meaning a Windows 7 system ramps to Turbo Boost faster, resulting in the better performance and higher power consumption. At idle, the previously-discussed features enable theWindows 7 config to dip below the idle power draw of the Vista machine.

Source

It appears that some are complaining about it's use and how they believe it effects performance. There is a way to disable it but I'm not sure if that will disable EIST (for example) if you make the change. Here's the gist of it:

I have tested the above on a clean Windows 7 x64 / Clean 8.5.1 install and it works perfectly - all Cores are available at all times and none are ever parked - the result = even CPU loads at all times regardless of loads.

In short, here is the better method from sky60234:-

- Go to Regedit

- Find this key:- " 0cc5b647-c1df-4637-891a-dec35c318583 "

- Within this key, there is a value called: " ValueMax "

- This value represents the % number of cores the system will park - the default 100% ie: all Cores are potentially park-able

- Change the value from 64 to 0 so the " ValueMin " and " ValueMax " are both zero

- You will have to find the key a few times and repeat the process for each time it is found - the number of instances will depend on the number of power profiles in your system [ in my DAW it was only found twice ]

- Do a full shutdown and power-off and cold-re-start

source

The change does not impact any of the Power Options settings for the CPU. Does anyone else have any additional information about Ideal Core, Core Parking and if it's a benefit to disable it as suggested above for even CPU loads? Is there any performance benefit disabling this? And, with this enabled did it have any adverse effects on benchmarking?

 

[EastCoasthandle]

Wow, no one knows??

 

[AsRock]

Nope never heard any thing of it but sounds interesting.

 

[EastCoasthandle]

From what I've gathered so far is that if you are using Windows 7 you are automatically using Core Parking. Core Parking is in essence like EIST (of sorts) making it more ideal for notebooks, netbooks, etc not PC's per say. However, I've not seen much information about it to suggest any real performance benefits with it enabled vs disabled.

 

[AsRock]

From what I've gathered so far is that if you are using Windows 7 you are automatically using Core Parking. Core Parking is in essence like EIST (of sorts) making it more ideal for notebooks, netbooks, etc not PC's per say. However, I've not seen much information about it to suggest any real performance benefits with it enabled vs disabled.

Enabling and disabling CPU cores just makes me think it just saves power is all. I'll try it tomorrow just for shits and giggles.

Well as long as i can get it to work in Vista but don't see why not.

 

[EastCoasthandle]

I'll do the same just to see...

 

[AsRock]

No luck here 0cc5b647-c1df-4637-891a-dec35c318583 don't exist in Vista.

 

[EastCoasthandle]

Hmm, I didn't check the same programs before but CPU Usage History appears to showing a lot more red activity (kernal mode) then before. And both cores seem to mirror each other when loaded. I don't recall that in the games I'ved played in the past nor watching videos. However, I really didn't pay much attention, lol.

 

[human_error]

win7 was given a lot more core optimisations over vista (one of the reasons i upgraded to win7 actually).

The most significant part comes with load balancing over the cores - it will obviously try to balance the load depending on if you are in a power saving or performance mode but there is extra logic where if you have a HT/SMT enabled processor (pentium4, atom, i5/i7) then it will avoid dumping threads onto a HT "core" unless the other cores are too busy or if the thread is marked as optimised for HT (meaning it wouldn't get the performance penalty other threads can get if running on HT cores). This works perfectly on my i7 rig - my real cores are used loads and it needs all my real cores to be 15-20% use before the computer puts anything which uses more than 3% of a core's resources onto a HT core, which means i get 0 performance loss running my rig with HT enabled (when bloomfield i7s originally launched you'd get a performance hit in some games/apps with HT on as important threads would be dumped into HT cores sharing logic with a busy real core, as opposed to one of the 3 other real cores doing nothing).

The extra power usage could be because the power mode chosen was not a power save option - in performance mode win7 should balance threads across cores so that the threads can run as fast as possible - it will obviously still chose a core which is in use if there would be no performance penalty but it will try and get the process completed as fast as possible. In power saving mode the logic should be that it will load one core to a medium load level before looking at other cores to use.

tl;dr version:

the most performance gains from the win7 core optimisations are for users with intel procs with HT/SMT or procs with deep sleep states for individual cores, otherwise you won't be noticing much.

 

[95Viper]

You can use Resource monitor in Windows 7 to see if any are parked. In Performance mode there should be no core parking, but in the power saving mode there will be.

This may explain better than I. http://blogs.technet.com/askperf/archive/2009/10/03/windows-7-windows-server-2008-r2-core-parking-intelligent-timer-tick-timer-coalescing.aspx

 

[EastCoasthandle]

OK, here is my experience after 24 hours of use this:

With valueMax set to 0 (from 100) I noticed mirrored CPU loading. Meaning they were both loaded equally when gaming. When valueMax is set to 100 (from 0) which is the default setting I noticed that one core was slightly more used then another. Using Resource Monitor for your CPUs or just Window Task Manger's CPU Usage History will show the same thing.

I didn't do any performance comparisons yet because I actually wanted to see if there was a difference. Also, a good CPU bound benchmark test would be the best route to see if there is any difference.

 

[Completely Bonkers]

ECH, this is interesting... but not sure yet on the benefits. Keep posting your results. Perhaps update your OP with criteria to what/which it applies.

 

[SystemViper]

yea, i am a little skeptical of any body other then me playing with my cores!

happy holidayzzzzzzzzzzzzzzzzzzzzzzzzz

 

[EastCoasthandle]

Windows 7 (check System Specs for details)

Core Parking (on by default)
3DMark06 (because it's cpu limited)
CPU Score: 3327
HDR/SM 3.0: 8317
SM 2.0: 6770
15847 (790/1100 video card)

Core Parking Disabled
3DMark06 (because it's CPU limited)
CPU Score: 3350
HDR/SM 3.0: 8338
SM 2.0: 6915
16002 (790/1100 video card)

Interesting results but more testing is needed...